Maker Pro

Desktop Case lights using WS2812B Leds.

Case lights using WS2812 led strips and a PIC18f14k50 combined with some extra logic.

Project Log

Hi this is my first project log on this site. I will be updating as I find time.

A few months back I bought some WS2812B led strips off ebay to play with but I haven't gotten around to working with them before now.
The microcontrollers I had around was some PIC18f14k50.

So what I first tried was the method of using fast clocked Spi to emulate the 1-wire protocol for the ws2812 leds. but it ended in failure and I could not get any consistent data transfer.
So I found a document from Microchip which explains how to use the CLC module they have on some of their PIC16f chips to interface with WS2811 led drivers.
link :

this document gave me an idea about how to create some custom external logic to fit the protocol.
After reading the document I figured out logically what I would have out would be:
Q = ( !SDO&CLK&PWM ) || ( SDO&CLK )

So my first attemt I build a circuit that did this with only logic gates:

And this seemed to work pretty good when I got the uC to run with some code and adjusted the PWM. But I figured out that it was a waste to have 3 IC's to do all this so I figured out I'd use some transistor logic instead to do some of the functions.

After a bit in thinking and simulating I had this circuit instead:

By switching out the inverter and OR gate I saved a lot of space.
The reason I did make a darlington connection at the end was due to the "or" gate being slow. I figure I could just have decreased the value on R6. but it works so I suppose it's fine.

Now I've been working on the program.

in my first prototype I was running it on a 48Mhz crystal. (well that was what I though)
But I figured out later that the PIC couldn't drive that kind of crystals and was only running at 20-30Mhz or somthing like that. not sure how to properly mesure that so I figured out I'd use the internal oscillator.
The PIC has the option to run it's internal 8Mhz osc though PLL I could gain a fairly stable 32Mhz cpu clock.
Since the CPU's arcitecture uses 4 clock pulses each instruction we can consider that a stable 12Mhz for all the internal pereferials.

This means that the internal instruction clock runs at 12Mhz. And that's where timers get their clock.

To drive the MSSP module I use the option of TMR2 / 2. this mean it will clock out a bit every secound time timer 2 resets.

Timer 2 is also the timer which generates the pwm. that's how we can keep those signals in sync. to get the right timing for the high pulse I adjusted the PR2 (the timer resets when it's value matches this register) until I had a 600ns high pulse.

After that I adjusted the pwm Duty cycle until I had a 400ns width on the signal. this makes sure that the "low" signal is only 400ns. See the document for further details on how the signal mixing works.

Since then I've been working on the code.

I will post the code soon. I need to copy it over to my new windows install.

The "best" working code has a "fade into all colors" function, a strobe and a change color + strobe, but it updates all the the leds to the same color.
But my most recent one is more focused on per-pixel animation and pumps out data for each pixel. Haven't made animations for this one yet though.

I did mount the led strips in my case and had it running the old code:

But in the process I managed to puncture my radiator and fry my PSU. (literally shorted the primary side of the powersupply, I can see it because the current sensing resistor and the fuse is pulverized)
So atm I'm without a desktop pc.
(that's what I get for being inpatient I suppose.)
  • Like
Reactions: Keng and Viktory2k1

Item information

Added by
Last update

More in Project Logs

Share this item